Refund policies for defective datasets

Introduction

Refund policies for defective datasets concern the legal rights and remedies available when a person or organisation purchases or licenses datasets that fail to meet agreed standards. Datasets are widely used in artificial intelligence, biotechnology, healthcare, scientific research, financial services and commercial decision-making. Their accuracy, completeness, relevance and reliability are essential for achieving the intended purpose.

A dataset may be considered defective when it contains inaccurate information, missing records, duplicated entries, outdated observations, incorrect classifications, corrupted files or improperly collected data. Defects may also arise where datasets lack promised metadata, contain unauthorised personal information or do not comply with contractual specifications. Refund policies establish how customers can seek repayment, replacement, correction or compensation when such problems occur.

Meaning and types of defective datasets

A defective dataset is a collection of digital information that does not satisfy the quality, functionality or legal requirements agreed upon between the supplier and purchaser. The defect may affect the entire dataset or only particular portions.

Common categories include:

Accuracy defects: Incorrect values, misleading labels or unreliable observations.

Completeness defects: Missing records, incomplete fields or absent data categories.

Technical defects: Corrupted files, incompatible formats or inaccessible databases.

Timeliness defects: Outdated information supplied despite a contractual requirement for current data.

Legal defects: Data collected or supplied without necessary permissions or in violation of applicable privacy requirements.

Representativeness defects: Data that fails to reflect the population or research group promised by the supplier.

In biotechnology, defective datasets can seriously affect genetic analysis, clinical research, molecular diagnostics and artificial intelligence models. Therefore, refund policies should distinguish between minor errors that can be corrected and substantial defects that make the dataset unsuitable for its intended purpose.

Legal framework in India

The Indian Contract Act, 1872, provides the principal contractual framework for disputes involving defective datasets. Section 37 requires parties to perform their contractual obligations. If a supplier fails to provide data conforming to agreed specifications, the purchaser may claim that the supplier has breached the contract.

Section 39 addresses situations where a party refuses to perform, or disables itself from performing, its contractual promise. Depending on the circumstances, the affected party may be entitled to terminate the contract.

Section 73 permits compensation for losses naturally arising from a breach of contract or those which the parties knew were likely to result from the breach. Section 74 governs compensation where a contract specifies a sum payable upon breach, subject to reasonable compensation not exceeding the amount stipulated.

The Consumer Protection Act, 2019, may also apply when datasets or related digital services are purchased by consumers for consideration and the transaction falls within the statutory definition of consumer services or goods. Section 2(11) defines deficiency in service, while Section 2(47) addresses unfair trade practices. Depending on the facts, consumer remedies may include removal of deficiencies, refund of consideration and compensation.

The Information Technology Act, 2000, and the Digital Personal Data Protection Act, 2023, may become relevant where defective datasets involve unauthorised access, unlawful disclosure or improper processing of personal data. A refund does not eliminate separate legal responsibilities arising from privacy or data-security violations.

Essential elements of a refund policy

An effective refund policy should clearly identify the circumstances in which a purchaser can request repayment. It should specify measurable quality standards, delivery requirements and procedures for reporting defects.

Important elements include:

Quality specifications: The contract should define acceptable levels of accuracy, completeness, duplication and consistency.

Inspection period: Purchasers should receive a reasonable period to examine the dataset and identify defects.

Evidence requirements: Complaints may be supported by validation reports, sample checks, technical assessments or independent audits.

Correction and replacement: Suppliers may be allowed to correct or replace defective data within a specified period.

Refund entitlement: A full or proportionate refund may be appropriate where defects are substantial or cannot be corrected.

Data deletion: Where a refund follows termination, the parties should address the return or deletion of dataset copies, subject to lawful retention requirements.

A refund clause should not unfairly exclude liability for deliberate misrepresentation, fraud or obligations that cannot legally be excluded.

Important case laws

1. M.C. Chacko v. State Bank of Travancore (1969) 2 SCC 343

The Supreme Court discussed the principle of privity of contract, under which contractual rights and obligations generally operate between the contracting parties. This principle is relevant to dataset refund disputes because the purchaser ordinarily enforces refund rights against the supplier with whom the contract was made. Third parties who merely use the dataset may not automatically acquire contractual refund rights.

2. Fateh Chand v. Balkishan Das, AIR 1963 SC 1405

The Supreme Court explained the principles governing compensation under Section 74 of the Indian Contract Act. It held that the court awards reasonable compensation rather than automatically enforcing the amount stated in a contractual penalty clause. This principle is relevant when dataset agreements prescribe fixed deductions, cancellation charges or compensation for defective delivery.

3. Kailash Nath Associates v. Delhi Development Authority (2015) 4 SCC 136

The Supreme Court clarified that compensation under Section 74 must be reasonable and that the existence of a breach is essential. The judgment is relevant when determining whether a supplier or purchaser can enforce predetermined financial consequences following a dataset contract dispute.

4. Indian Medical Association v. V.P. Shantha (1995) 6 SCC 651

The Supreme Court explained when medical services fall within the scope of consumer protection law. Although the case did not concern datasets, its principles may assist in determining whether a paid healthcare-related data service qualifies as a consumer service, depending on the nature of the transaction.

Challenges in implementing refund policies

One major difficulty is determining whether a dataset is genuinely defective or simply unsuitable for a purchaser's particular research model. A supplier may have delivered technically accurate data, but the purchaser's algorithm or methodology may produce unsatisfactory results.

Another challenge concerns datasets that have already been integrated into machine-learning systems or research databases. Returning the original files may not remove their influence from trained models or derived analytical outputs. Contracts should therefore establish rules for deletion, retraining, restricted reuse and retention of derived information.

Cross-border transactions create additional difficulties because different jurisdictions may apply different contractual, consumer-protection and privacy requirements. Parties should clearly identify governing law, dispute-resolution procedures and applicable refund timelines.

Conclusion

Refund policies for defective datasets are essential for maintaining transparency, accountability and confidence in digital information markets. Under Indian law, the Contract Act, 1872, provides remedies for contractual breaches, while the Consumer Protection Act, 2019, may offer additional protection where its requirements are satisfied. Effective policies should establish clear quality standards, inspection periods, correction procedures and refund conditions. In biotechnology and scientific research, these safeguards are particularly important because defective datasets can undermine research accuracy, healthcare decisions and technological development.